INDEX

Explanations

phrases indicating the importance or contribution of specific elements or actions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

not only

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Gweler

-0.47

ⓧ

-0.46

Климат

-0.45

 يتيمه

-0.43

Datuak

-0.42

featureID

-0.41

SharedDtor

-0.41

fromnode

-0.40

RegressionTest

-0.40

الدراسه

-0.39

POSITIVE LOGITS

 surla

0.49

styleType

0.43

sho

0.39

 syno

0.39

 rapidly

0.38

maketitle

0.37

̈́

0.36

éez

0.36

Vidite

0.36

GOTREF

0.36

Activations Density 0.035%

phrases indicating the importance or contribution of specific elements or actions

not only

No Comments

No Known Activations